Longitudinal Association of Inflammatory Diets on Cognition in Older Adults: Insights from the Oral–Gut–Brain Axis

ABSTRACT

Background

The oral–gut–brain axis is known to influence dementia development, but the interactions between an inflammatory diet, inflammatory conditions such as periodontitis and Helicobacter pylori (Hp) infection, and their effects on cognitive function remain unclear. This study aims to investigate how periodontitis and Hp infection affect the association between an inflammatory diet and cognitive performance.

Methods

This prospective cohort study was conducted from 2011 to 2019, involving community‐dwelling older adults (≥ 65 years old) without dementia, recruited between 2011 and 2013 (N = 511). At baseline, dietary inflammatory potential was assessed by the Empirical Dietary Inflammatory Index (EDII) using a 44‐item semi‐quantitative food frequency questionnaire. Periodontal status was evaluated by a dentist, and Hp immunoglobulin G levels were measured. Global and domain‐specific cognitive functions were assessed at baseline and during three biennial follow‐ups. A generalized linear mixed model was used to examine the association between the EDII and cognitive function, adjusting for important covariates. Stratified analyses were further conducted by periodontal status and Hp seropositivity, respectively.

Results

One unit increase in the EDII, indicating a more pro‐inflammatory diet, was associated with poor memory performance (Logical Memory‐immediate free recall: $\widehat{\beta }$ = −0.61, 95% confidence interval (CI) = −1.02 to −0.21; delayed free recall: $\widehat{\beta }$ = −0.64, 95% CI = −1.06 to −0.22). This association was more pronounced among participants with periodontitis and Hp seropositivity ($\widehat{\beta }$ = −1.15 to −0.82). Significant interactions were found between the EDII and periodontitis in the memory domain (p _interaction = 0.01).

Conclusions

The association between an inflammatory diet and memory may be more pronounced in older adults with periodontitis and Hp seropositivity. These findings support the relevance of healthy eating, oral health maintenance, and Hp management in maintaining cognitive health.

Summary.

• Key points

  • ○
    An inflammatory diet is associated with cognitive performance, particularly in the memory domain in older adults.
  • ○
    The association between an inflammatory diet and memory appeared more pronounced in older adults with periodontitis and Helicobacter pylori seropositivity.
• Why does this matter?

  • ○
    A holistic approach involving healthy eating, oral health maintenance, and H. pylori management may play a role in supporting cognitive health.

1. Introduction

The number of dementia cases is estimated to rise from 57.4 million in 2019 to 152.8 million by 2050, presenting a significant global challenge [1]. Alzheimer's disease (AD), the most common type of dementia, accounts for 60%–80% of cases and is the seventh leading cause of death in the United States [2]. Taiwan has a rapidly aging population, with dementia prevalence reported at 8% and mild cognitive impairment prevalence at 18.8% [3]. Older adults with dementia often experience prolonged disability, placing substantial strain on families, healthcare, and social services [2].

Currently, no effective treatment exists for dementia. In addition to the hallmarks of AD, which include the amyloid cascade and tau propagation, the neuroinflammation hypothesis has garnered particular attention [4, 5]. Neuroinflammation is believed to begin decades before AD onset, offering a potential window for early intervention [4]. Since diet may modulate inflammation, investigating inflammatory diets and cognition could provide insights into early dementia prevention [6].

Various methods have been utilized to derive inflammatory dietary patterns [7], enabling researchers to evaluate their associations with cognition. Some studies use reduced rank regression (RRR) to identify inflammatory dietary patterns linked to biomarkers such as C‐reactive protein (CRP) and interleukin‐6 (IL‐6), which are associated with cognitive decline and reduced brain volume [8, 9]. Other studies using a priori methods find that a higher nutrient‐based Dietary Inflammatory Index (DII) is associated with cognitive impairment [10, 11, 12, 13, 14, 15], dementia [16, 17], and specific brain volume changes [17], except for one study showing no association [18]. However, most studies were conducted in Western countries, except one in Korea [11]. Given dietary and lifestyle differences, investigating these associations across diverse cultural contexts is essential.

Another relevant a priori inflammatory dietary index is the Empirical Dietary Inflammatory Index (EDII), developed using RRR with three biomarkers (IL‐6, CRP, and tumor necrosis factor‐α receptor 2) in the Nurses' Health Study (NHS) [19]. It has been further validated in men and multiethnic populations [19, 20]. Due to its food‐based approach, the EDII directly reflects individuals' perceived diets, making it particularly suitable for public health promotion.

Recent studies have underscored the disruption of the oral–gut–brain axis in dementia development [5, 21]. Periodontitis and Helicobacter pylori (Hp) infection are associated with oral and gut dysbiosis, systemic inflammation, and cognitive impairment or decline [22, 23, 24]. Diet may further induce dysbiosis in these conditions [25, 26], potentially amplifying systemic inflammation and its downstream effects on cognition. However, research on the interrelationships among diet, periodontitis, and Hp infection on cognition remains scarce. Existing studies have examined diet and cognition in periodontitis patients without accounting for Hp infection [27] or have focused on inflammatory diet and Hp infection in relation to mortality rather than cognition [28], limiting understanding of their effect on diet‐related cognitive impairment.

To address this gap, we conducted a prospective cohort study to investigate the association between an inflammatory diet, as measured by the EDII, and global and domain‐specific cognitive functions in community‐dwelling older adults. We further examined whether periodontitis and Hp infection affect the association between an inflammatory diet and cognitive function.

2. Methods

2.1. Study Population and Design

The Taiwan Initiative for Geriatric Epidemiological Research is an ongoing prospective cohort study (2011–present) involving 605 community‐dwelling older adults aged 65 years or older recruited from the annual Senior Health Checkup Program at the National Taiwan University Hospital (NTUH) at baseline (2011–2013). Participants with the following criteria were excluded at baseline: (1) clinically diagnosed dementia; (2) use of medications for AD; (3) suspected dementia, as indicated by a Montreal Cognitive Assessment–Taiwanese version (MoCA–T) score ≤ 21; (4) brain tumor (≥ 3 cm) identified by magnetic resonance imaging; (5) history of stroke or head trauma; (6) missing food frequency questionnaire (FFQ) data; (7) implausible total energy intakes (< 650 kcal/day or > 3450 kcal/day); and (8) vegetarians, as the FFQ was not developed for this subpopulation. A total of 511 participants were included and followed biennially, with 359 remaining at the 6‐year follow‐up (Figure S1). Among them, missing cognitive test data were 0.8% at the 4‐year follow‐up. Missing data for other covariates ranged from 0% to 6% at each time point. This study was approved by the NTUH Research Ethics Committee, and informed consent was obtained from each participant before enrollment.

2.2. Dietary Assessment and EDII Calculation

At baseline (2011–2013), a validated 44‐item semi‐quantitative FFQ was administered through face‐to‐face interviews by trained research assistants to assess participants' usual dietary intake over the previous year [29]. The FFQ included dairy products, eggs, meat, poultry, fish, seafood, offal, soybean products, vegetables, fruits, grains (refined and whole), tubers, sugar‐sweetened beverages, sweet bread, pickled vegetables, roasted meats, fermented foods, fried foods, coffee, tea, alcohol, supplements, and cooking oils. Participants reported intake frequency using a seven‐category scale, ranging from “never or less than once per month” to “twice or more per day,” based on representative portion sizes. Total energy intake was estimated using Taiwan's Food and Drug Administration database [30].

The EDII was used to assess the dietary inflammatory potential. Due to the uncommon consumption or unavailability of certain food items in our FFQ, pizza, processed meat, snacks, fruit juice, and low‐energy beverages were excluded from the EDII calculation. Thus, 13 of the 18 food groups were included, comprising seven pro‐inflammatory food groups (red meat, organ meat, other fish, other vegetables, refined grains, high‐energy beverages, and tomatoes) and six anti‐inflammatory food groups (beer, wine, tea, coffee, dark‐yellow vegetables, and leafy green vegetables) (Table S1).

We used intake frequency and portion size obtained from our FFQ to calculate the daily serving size of each food group, which was then converted to the portion sizes defined by the original EDII (Table S1). The defined portion size for each food group was multiplied by its corresponding weight from the original EDII [19]. The EDII scores were obtained by summing the weighted food group intakes and subsequently dividing by 1000 to reduce magnitude for easier interpretation. A higher EDII score indicated a more pro‐inflammatory diet. We examined EDII both as a continuous variable and as a categorical variable, classifying participants into tertiles representing low (T1), moderate (T2), and high (T3) inflammatory diet groups.

2.3. Assessment of Periodontal Status

At baseline, a qualified dentist performed oral examinations. The dental exams were conducted indoors under artificial lighting, with participants examined in mobile dental chairs in a recumbent position. Periodontal status was classified into three categories: healthy periodontium, presence of calculus/gingivitis, and presence of periodontitis.

2.4. Cognitive Assessments

Trained research assistants administered a series of neuropsychological tests to assess global and domain‐specific cognitive functions, including (1) global cognition, assessed by the MoCA–T (≥ 24: normal cognition, < 24: cognitive impairment, ≤ 21: suspect dementia) [31], (2) memory domain, evaluated by Logical Memory‐immediate and delayed recall (both thematic and free) from the Wechsler memory scale—third edition (WMS‐III) [32], (3) attention domain, assessed by forward and backward digit span tasks from the WMS‐III; (4) executive function, assessed using the Trail Making Tests A and B; and (5) verbal fluency, assessed by naming fish, vegetables, and fruit within 1 min per category, with the results summed up. Trail Making Test scores were multiplied by (−1) to align scoring direction across domains. Cognitive test scores were standardized using baseline means and standard deviations to produce Z‐scores for comparability. Cognitive assessments were conducted at baseline and biennially. To account for practice effects, models were adjusted for the number of prior cognitive tests.

2.5. Biomarkers

At baseline, each participant underwent venous blood collection in the morning after fasting for 8 h. Blood samples were stored at −80°C until analysis. The Hp immunoglobulin G test was performed using a commercial enzyme immunoassay (IMMULITE 2000, Siemens, Germany), with an antibody titer of ≥ 1.1 U/mL defined as Hp seropositivity [23]. The apolipoprotein E (APOE) e4 status was determined by genotyping two single‐nucleotide polymorphisms (rs429358 and rs7412) using TaqMan Genomic Assays and an ABI 7900HT real‐time polymerase chain reaction system (Applied Biosystems Inc., Foster City, CA, USA).

2.6. Covariates

To control for potential confounders, we collected the following covariates: participants' demographic and clinical characteristics (age, sex, education years, annual family income, smoking status, medical history, and medication or supplement use), gathered by a trained interviewer. Physical activity during the previous week was assessed using the short version of the International Physical Activity Questionnaire [33]. Depressive symptoms were defined as a Center for Epidemiologic Studies Depression Scale score ≥ 16, a history of depression, or antidepressant use [34]. An inflammation‐related comorbidity score was calculated by summing the presence (1) or absence (0) of hypertension, diabetes, hyperlipidemia, and arthritis. Anti‐inflammatory medication use includes nonsteroidal anti‐inflammatory drugs and steroids. Body mass index (BMI) was calculated as weight (kg) divided by height squared (m²).

2.7. Statistical Analysis

To compare the distributions of each variable across the EDII groups, the Kruskal–Wallis test was used for continuous variables, while the chi‐square test was employed for categorical variables.

For multivariable analyses, generalized linear mixed models (GLMMs) were used to estimate adjusted odds ratios and 95% confidence intervals for the association between the EDII (treated as a continuous variable or tertile) and the risk of cognitive impairment (MoCA–T < 24). The GLMMs were also used to obtain the regression coefficient β to evaluate the performance of global or domain‐specific cognition for continuous or tertiled EDII over time. Time interaction terms were included for variables available only at baseline. All models were adjusted for sex, baseline age, education years, APOE e4 status, total caloric intake, periodontal status, Hp seropositivity, and time‐varying covariates with repeated measures, including depressive symptoms, physical activity, BMI, inflammation‐related comorbidity score, annual family income, practice effects, and years from baseline (i.e., follow‐up time).

Stratified analyses by periodontal status, Hp seropositivity, and sex were performed for significant associations between an inflammatory diet and cognition to identify susceptible subpopulations. Subsequently, interactions between the EDII and periodontal status, Hp seropositivity, and sex were tested by including their product terms in the model.

Furthermore, we performed sensitivity analyses. Since the FFQ was only administered at baseline, we conducted a sensitivity analysis excluding participants who reported a “change in eating habits” based on a question from the Social Readjustment Rating Scale [35], collected during the 4‐year follow‐up.

All statistical analyses were conducted using SAS version 9.4 (SAS Institute Inc., Cary, NC, USA). Statistical significance was set at p < 0.05.

3. Results

3.1. Characteristics of the Study Participants

The median follow‐up time was 5.7 years. At baseline, participants had a mean age of 72.6 years, and all were of Taiwanese ethnicity. The distribution by sex was comparable, with women comprising 53% of the study population. Approximately one‐third of the participants were classified as overweight or obese (BMI ≥ 25 kg/m²). One‐fourth (25%) of the participants were diagnosed with periodontitis, and 45% showed Hp seropositivity (≥ 1.1 U/mL). Total energy intake differed significantly across the tertiles of the EDII (Table 1). Participants lost to follow‐up differed from those who remained, particularly in age, education, cognitive function, and several health conditions (Table S2). Transitions in cognitive status over time are illustrated in Figure S2.

TABLE 1.

Characteristics of the study population by tertiles of the Empirical Dietary Inflammatory Index (EDII) at baseline (2011–2013).

Variables	EDII				p
	Overall	Low (T1)	Moderate (T2)	High (T3)
	(n = 511)	(n = 170)	(n = 171)	(n = 170)
	Mean (SD)
EDII	0.0 (0.2)	−0.2 (0.1)	0.0 (0.0)	0.3 (0.1)	< 0.001
Age (years)	72.6 (5.3)	72.6 (5.3)	72.7 (5.1)	72.6 (5.4)	0.95
Education (years)	13.8 (3.6)	14.2 (3.5)	13.5 (3.9)	13.7 (3.4)	0.19
BMI (kg/m2)	23.9 (2.9)	23.8 (3.0)	23.6 (2.6)	24.2 (3.2)	0.31
Inflammation‐related comorbidity score	1.4 (0.9)	1.4 (0.9)	1.4 (1.0)	1.4 (0.9)	0.98
Physical activity (MET‐min/week)	1736.5 (1459.3)	1839.6 (1461.1)	1625.9 (1258.8)	1744.0 (1643.3)	0.38
Total energy intake (kcal/day)	1664.2 (403.5)	1574.8 (396.4)	1583.0 (367.7)	1835.3 (392.5)	< 0.001
MoCA–T	26.9 (2.2)	27.1 (2.2)	27.1 (2.2)	26.7 (2.1)	0.12

Variables	EDII				p
	Overall	Low (T1)	Moderate (T2)	High (T3)
	(n = 511)	(n = 170)	(n = 171)	(n = 170)
	n (%)
Women	271 (53)	88 (52)	101 (59)	82 (48)	0.12
APOE e4 carriers	79 (16)	30 (18)	24 (14)	25 (15)	0.66
Household income > 1000k (TWD per year)	278 (58)	85 (52)	103 (64)	90 (57)	0.12
Ever smoker	82 (16)	30 (18)	20 (12)	32 (19)	0.16
Supplement users	403 (79)	133 (78)	140 (82)	130 (76)	0.46
Anti‐inflammatory medication users	149 (29)	44 (26)	53 (31)	52 (31)	0.51
Depressive symptoms	52 (10)	17 (10)	17 (10)	18 (11)	0.35
Hypertension	299 (59)	99 (58)	97 (57)	103 (61)	0.77
Diabetes mellitus	79 (15)	23 (14)	32 (19)	24 (14)	0.35
Periodontal status					0.50
Health	76 (15)	29 (17)	22 (13)	25 (15)
Gingivitis	302 (60)	102 (60)	97 (57)	103 (62)
Periodontitis	129 (25)	39 (23)	51 (30)	39 (23)
Hp seropositivity	229 (45)	73 (43)	74 (43)	82 (48)	0.55

Note: Numbers in bold indicate statistically significant findings (p < 0.05).

Abbreviations: APOE, apolipoprotein E; BMI, body mass index; Hp, Helicobacter pylori ; MET, metabolic equivalent of task; MoCA–T, Montreal Cognitive Assessment–Taiwanese version; SD, standard deviation; T, tertile; TWD, Taiwan dollar.

3.2. Association Between an Inflammatory Diet and Cognition

Table 2 demonstrates the longitudinal association between an inflammatory diet (assessed by EDII) and cognitive performance over time. One unit increase in the EDII (i.e., a more inflammatory diet) was associated with poor memory performance (Logical Memory‐immediate free recall: $\widehat{\beta }$ = −0.61, p = 0.003; delayed free recall: $\widehat{\beta }$ = −0.64, p = 0.003). Similar findings were found when we tertiled the EDII scores into low‐, moderate‐, and high‐inflammatory diet groups. A high‐inflammatory diet was associated with poor memory performance ($\widehat{\beta }$ = −0.36 to −0.34, p _trend = 0.001–0.002) compared with a low‐inflammatory diet. Additionally, a moderately inflammatory diet (T2) was significantly associated with better verbal fluency ($\widehat{\beta }$ = 0.20) compared with a low‐inflammatory diet.

TABLE 2.

Associations between the Empirical Dietary Inflammatory Index (EDII) and cognition over 8 years (2011–2019).

	Global cognition (MoCA‐T)		Logical memory				Trail making		Digit span		Verbal fluency
	Impaired a vs. normal	Continuous score	Immediate theme recall	Delayed theme recall	Immediate free recall	Delayed free recall	A	B	Forward	Backward
	aOR	$\widehat{\beta }$	$\widehat{\beta }$	$\widehat{\beta }$	$\widehat{\beta }$	$\widehat{\beta }$	$\widehat{\beta }$	$\widehat{\beta }$	$\widehat{\beta }$	$\widehat{\beta }$	$\widehat{\beta }$
	(95% CI)	(95% CI)	(95% CI)	(95% CI)	(95% CI)	(95% CI)	(95% CI)	(95% CI)	(95% CI)	(95% CI)	(95% CI)
Continuous EDII
EDII	2.24	−0.80	−0.26	−0.32	−0.61	−0.64	−0.24	−0.19	0.05	−0.09	−0.22
	(0.43, 11.73)	(−1.75, 0.16)	(−0.66, 0.15)	(−0.73, 0.08)	(−1.02, −0.21)	(−1.06, −0.22)	(−0.59, 0.10)	(−0.57, 0.19)	(−0.33, 0.42)	(−0.45, 0.27)	(−0.60, 0.16)
EDII × FU time	1.02	0.03	0.02	−0.01	−0.0002	−0.001	0.005	−0.03	−0.03	−0.02	−0.03
	(0.66, 1.58)	(−0.16, 0.21)	(−0.07, 0.10)	(−0.09, 0.07)	(−0.07, 0.07)	(−0.07, 0.07)	(−0.05, 0.06)	(−0.09, 0.02)	(−0.10, 0.04)	(−0.09, 0.05)	(−0.09, 0.02)
Tertiled EDII
T1	1.00	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.
T2	0.87	−0.04	−0.08	−0.07	−0.24	−0.28	−0.01	−0.11	0.07	−0.05	0.20
	(0.35, 2.19)	(−0.53, 0.44)	(−0.28, 0.13)	(−0.28, 0.14)	(−0.44, −0.03)	(−0.50, −0.07)	(−0.18, 0.17)	(−0.30, 0.09)	(−0.12, 0.26)	(−0.23, 0.14)	(0.01, 0.40)
T3	1.33	−0.38	−0.08	−0.12	−0.36	−0.34	−0.08	−0.10	0.04	−0.05	−0.08
	(0.56, 3.15)	(−0.88, 0.12)	(−0.30, 0.13)	(−0.33, 0.09)	(−0.57, −0.15)	(−0.56, −0.13)	(−0.26, 0.10)	(−0.29, 0.10)	(−0.16, 0.23)	(−0.24, 0.14)	(−0.27, 0.12)
T2 × FU time	1.17	−0.03	−0.03	−0.02	−0.03	−0.02	0.02	0.02	0.01	−0.001	−0.01
	(0.91, 1.51)	(−0.13, 0.07)	(−0.08, 0.01)	(−0.06, 0.02)	(−0.07, 0.003)	(−0.05, 0.02)	(−0.01, 0.05)	(−0.01, 0.05)	(−0.03, 0.04)	(−0.04, 0.04)	(−0.04, 0.01)
T3 × FU time	1.10	−0.04	−0.01	−0.02	0.003	−0.004	−0.004	−0.01	−0.01	−0.01	−0.02
	(0.86, 1.40)	(−0.14, 0.06)	(−0.05, 0.03)	(−0.06, 0.02)	(−0.03, 0.04)	(−0.04, 0.03)	(−0.03, 0.03)	(−0.04, 0.02)	(−0.05, 0.02)	(−0.05, 0.03)	(−0.05, 0.01)
p trend	0.50	0.14	0.42	0.42	0.001	0.002	0.41	0.34	0.68	0.61	0.52

Note: Generalized linear mixed models were used to explore the association of EDII (continuous or tertile) and EDII × FU time with cognition, adjusted for age, sex, education, APOE e4 status, depressive symptoms, physical activity, BMI, total calorie intake, inflammation‐related comorbidity score, income, Hp seropositivity, Hp seropositivity × FU time, periodontitis, periodontitis × FU time, practice effects, and FU time. Numbers in bold indicate significant statistical findings (p < 0.05).

Abbreviations: aOR, adjusted odds ratio; APOE, apolipoprotein E; BMI, body mass index; CI, confidence interval; FU, follow‐up; Hp, Helicobacter pylori ; MoCA–T, Montreal Cognitive Assessment–Taiwanese version; Ref., reference.

^a Impaired cognition was defined as MoCA–T < 24.

3.3. Stratified Analyses by the Presence of Periodontitis and Hp Seropositivity

Based on the significant findings in Table 2, we performed stratified analyses to explore the associations between the inflammatory diet (EDII) and cognition, considering the presence of periodontitis and Hp seropositivity (Figure 1 and Table S3).

FIGURE 1.

Forest plots of the associations between the Empirical Dietary Inflammatory index (EDII) and cognition over 8 years, stratified by the presence of periodontitis and Helicobacter pylori (Hp) seropositivity. (A) Associations stratified by the presence of periodontitis. (B) Associations stratified by Hp seropositivity. Forest plots display the effects on cognition across the study period. Generalized linear mixed models, with cognition as the dependent variable and EDII (continuous) and EDII (continuous) × FU time as independent variables, are depicted. Models were adjusted for age, sex, education, APOE e4 status, depressive symptoms, physical activity, BMI, total calorie intake, inflammation‐related comorbidity score, income, practice effects, and FU time. Additional adjustments included Hp seropositivity when stratifying by the presence of periodontitis; and periodontitis when stratifying by the presence of Hp seropositivity. Numbers in bold indicate significant findings (p < 0.05). APOE, apolipoprotein E; BMI, body mass index; CI, confidence interval; FU, follow‐up.

Significant interactions were observed between the EDII and periodontitis in the memory domain (p _interaction = 0.01), whereas no significant interactions were found for Hp seropositivity. After stratification, significant findings were observed in certain subpopulations. Among participants with periodontitis, one unit increase in the EDII was associated with poor memory performance (Logical Memory‐immediate free recall: $\widehat{\beta }$ = −0.97, p = 0.01; delayed free recall: $\widehat{\beta }$ = −1.15, p = 0.003). Similarly, among participants with Hp seropositivity, one unit increase in the EDII was associated with poor memory performance (Logical Memory‐immediate free recall: $\widehat{\beta }$ = −0.89, p = 0.01; delayed free recall: $\widehat{\beta }$ = −0.82, p = 0.01).

3.4. Stratified Analyses by Sex

Based on significant findings in Table 2, we further stratified the association between EDII and cognition by sex. No significant interaction was found (Table S4). After stratification, the associations between EDII and poor memory performance were more pronounced in men (Logical Memory‐immediate free recall: $\widehat{\beta }$ = −0.72, p = 0.02; delayed free recall: $\widehat{\beta }$ = −0.78, p = 0.01). No significant association was found in women.

3.5. Sensitivity Analysis

Among the 393 participants retained in the 4‐year follow‐up (2015–2017), 47 (12%) reported a “change in eating habits.” We repeated the analysis after excluding these participants and found that the association remained similar, with one unit increase in the EDII significantly associated with poor memory performance (Logical Memory‐immediate free recall: $\widehat{\beta }$ = −0.61, p = 0.01; delayed free recall: $\widehat{\beta }$ = −0.59, p = 0.01). Additionally, one unit increase in the EDII was significantly associated with poor global cognition (MoCA–T: $\widehat{\beta }$ = −1.01, p = 0.048) (Table S5).

Stratified analyses yielded similar results. One unit increase in the EDII was significantly associated with poor memory performance among participants with periodontitis and Hp seropositivity ($\widehat{\beta }$ = −1.21 to −0.79). Significant interactions were observed between EDII and periodontitis in the memory domain (p _interaction = 0.01), whereas no significant interaction was found for Hp seropositivity (Table S6).

4. Discussion

To our knowledge, this study newly explores the interrelationships among inflammatory diets, inflammatory conditions (including periodontitis and Hp infection), and cognition, grounded in the oral–gut–brain axis theory [5, 21]. We found that a more pro‐inflammatory diet, as assessed by the EDII, was associated with poor memory performance. Importantly, this association became more pronounced in older adults with periodontitis or Hp seropositivity. This study fills the knowledge gap regarding the association between an inflammatory diet and cognition based on the insight of the oral–gut–brain axis.

We found that an inflammatory diet was associated with impaired memory, a domain closely linked to AD [36], and this finding is consistent with previous studies [10, 13]. However, one study using an RRR‐identified inflammatory nutrient pattern reported associations with visuospatial deficits [8]. In contrast, the NHS, which utilized the original EDII, found an inverse association in older women, where a higher EDII (i.e., a more inflammatory diet) was associated with reduced declines in both global and verbal memory over 6 years [37]. These inconsistencies may be attributed to differences in sex (a more pronounced effect was found among older men in our study), ethnicity, education levels (the NHS included only nursing professionals), dietary assessment methods, and cognitive tests used.

The significant association between a moderately inflammatory diet and better verbal fluency may reflect domain‐specific cognitive vulnerability. EDII appears to have a stronger effect on memory‐related functions, as shown by significant associations in two related tests, while verbal fluency may be less susceptible to an inflammatory diet. Additionally, some pro‐inflammatory foods in the EDII (e.g., fatty fish and certain vegetables) contain bioactive compounds that may support cognition [38]. Further research is needed to clarify the role of specific dietary components in cognitive health.

Our findings suggest that the association of an inflammatory diet on cognition was more pronounced in participants with periodontitis. Interactions between periodontitis and an inflammatory diet were significantly associated with poorer memory performance. Previously, one cross‐sectional study reported an association between an unhealthy dietary pattern and a higher risk of cognitive decline in older adults with periodontitis [27], limiting causal inference. Another study suggested that an inflammatory diet may mediate the association between periodontitis and cognition [39]. However, the relationship between periodontitis and an inflammatory diet may be bidirectional rather than unidirectional.

Furthermore, we discovered that the association between an inflammatory diet and memory was more pronounced among participants with Hp seropositivity, a finding that has not been previously reported. Other studies exploring inflammatory diets and Hp infection focused on different outcomes. For example, a murine study revealed that Hp infection exacerbates central obesity and insulin resistance induced by a high‐fat diet [40]. Another study suggested that a pro‐inflammatory diet is associated with increased all‐cause mortality, especially in individuals with Hp infection [28]. Together with our results, these findings underscore the complex interplay between diet, Hp infection, and health outcomes, highlighting the need for further research on their combined effect on cognitive function.

Postulated mechanisms regarding an inflammatory diet and cognition are summarized below. Certain EDII‐defined foods (inflammatory diet) are associated with inflammation and cognition. For example, red and organ meats (high in saturated fat and cholesterol), refined grains, and sugary beverages (rich in simple carbohydrates) may contribute to systemic inflammation and cognitive impairment [41]. In contrast, beer, dark‐green vegetables (rich in B vitamins), grape‐derived wine (containing resveratrol), tea and coffee (with caffeine and polyphenols), and dark‐yellow vegetables (containing carotenes) exhibit anti‐inflammatory potential and cognitive benefits [38, 42]. Common Chinese cooking methods, such as pan‐ and stir‐frying, tend to increase the dietary advanced glycation end products in fish and red meat, promoting inflammation and subsequent cognitive decline [43].

Furthermore, diet may trigger microbiota dysbiosis, leading to systemic low‐grade inflammation and disrupting the gut‐brain axis [5, 21]. An unhealthy diet, such as the Western diet, disrupts microbiota homeostasis and compromises intestinal epithelial integrity [5]. This increases intestinal permeability, allowing lipopolysaccharides into circulation and triggering the CD14 and toll‐like receptor 4 pathway. Consequently, systemic inflammation damages the blood–brain barrier (BBB), exacerbating its vulnerability [44, 45].

Additionally, periodontal pathogens and Hp can migrate to the brain via the trigeminal nerve, oral–nasal–olfactory route, or bloodstream, altering BBB permeability and causing brain damage [46, 47]. They may also contribute to AD pathology by activating inflammatory and immune responses via bloodstream mediators [24, 46]. Unhealthy diets induce dysbiosis in individuals with periodontitis and Hp infection [25, 26], potentially synergizing to alter BBB permeability and worsen these adverse effects. Given the complex interrelationships among diet, microbiota, and cognition, further exploration is required.

This study has several strengths. To the best of our knowledge, this is the first longitudinal cohort study to examine the interrelationships among inflammatory dietary patterns, Hp infection, periodontitis, and cognitive function. Second, in addition to lifestyle and chronic disease profiles, we further adjusted for the genetic risk factor for AD, the APOE e4 status, which has been under‐explored in previous studies investigating the relationship between inflammatory diet and cognition. Third, this study utilized multiple neuropsychological tests to assess four cognitive domains, allowing us to identify early cognitive changes in older adults without dementia. Additionally, this cohort study incorporated four repeated cognitive measures on older adults without dementia, providing valuable insights into the mechanisms underlying the relationship between inflammatory dietary patterns and cognition.

Our study has some limitations. First, our EDII included 13 out of 18 food items due to dietary differences. While the original EDII was validated in a multiethnic U.S. cohort (2.4% Asian or Pacific Islander) [20], modified versions using 10 to 16 food items have shown inconsistent associations with different health outcomes [48, 49, 50]. These discrepancies may reflect differences in dietary patterns, scoring strategies, and outcomes examined. Nonetheless, our findings suggest that an inflammatory diet is associated with poorer memory performance but better verbal fluency, reflecting inconsistencies across cognitive domains. Further studies should develop a locally adapted index of dietary inflammation incorporating Taiwanese dietary patterns and inflammatory biomarkers. Second, EDII was assessed using a baseline FFQ, which may not fully capture long‐term inflammatory dietary exposure. A sensitivity analysis excluding participants with dietary changes in this study yielded consistent findings. As dietary choices may change with health status, we adjusted for comorbidities over time to minimize their confounding effect. Since the FFQ relies on recall, cognitive decline over time may affect validity. However, excluding participants with dementia at baseline (i.e., prevalent cases) helps reduce survivor bias and improves temporality, as exposure information for incident cases was collected after recruitment. Measurement errors from the short FFQ are also possible, but the prospective design likely led to non‐differential misclassification, biasing results toward the null. Third, Table 1 presents crude descriptive analyses without adjusting for confounders, which may lead to biased or unexpected results in this heterogeneous population. This underscores the importance of multivariable regression in providing more accurate estimates. Fourth, participants were recruited from a Senior Health Checkup Program at a tertiary hospital in Taipei. They were more health‐conscious, had higher socioeconomic status, and were exclusively East Asians. While this population strengthens internal validity, it may limit the generalizability of our findings to Western populations. Fifth, selective attrition—where participants lost to follow‐up are more likely to have poorer cognitive outcomes—has been observed in older populations and may lead to underestimation of associations. Therefore, our findings should be interpreted with caution. Finally, this study did not assess periodontal pathogens, gut microbiota, or inflammatory biomarkers. In addition, recall bias inherent in dietary questionnaires may be a limitation. Future studies incorporating these measures will help elucidate underlying mechanisms.

In conclusion, a holistic approach, including healthy eating, oral health maintenance, and Hp management, may play a role in supporting cognitive health. Further research is needed to clarify the links between diet, microbiota, and cognition to better understand underlying mechanisms.

Author Contributions

Yi‐Chun Chou, Tina H.T. Chiu, Jen‐Hau Chen, and Yen‐Ching Chen designed the research. Jen‐Hau Chen and Yen‐Ching Chen conducted the research. Yi‐Chun Chou, Ming‐Lun Han, Tina H.T. Chiu, Meei‐Shyuan Lee, and Jeng‐Min Chiou analyzed the data. Yi‐Chun Chou, Jen‐Hau Chen, and Yen‐Ching Chen wrote the paper. Jen‐Hau Chen and Yen‐Ching Chen acquired funding. Yen‐Ching Chen was the primary person responsible for the final content. All authors have read and approved the final manuscript.

Disclosure

The authors have nothing to report.

Conflicts of Interest

The authors declare no conflicts of interest.

Linked Article

This publication is linked to a related editorial by Peter M. Abadir. To view this article, visit https://doi.org/10.1111/jgs.19623.

Supporting information

Figure S1. Flowchart of the study participants.

Figure S2. Transitions in cognitive status, loss to follow‐up, and mortality over time.

Table S1. Food items, portion sizes, and weights in the original and our modified Empirical Dietary Inflammatory Index (EDII).

Table S2. Baseline characteristics of participants by follow‐up status.

Table S3. Associations between Empirical Dietary Inflammatory Index (EDII) and cognition over 8 years, stratified by the presence of periodontitis and Helicobacter pylori (Hp) seropositivity.

Table S4. Associations between the Empirical Dietary Inflammatory Index (EDII) and cognition over 8 years, stratified by sex.

Table S5. Sensitivity analyses: association between the Empirical Dietary Inflammatory Index (EDII) and cognition over 8 years in participants with unchanged eating habits.

Table S6. Sensitivity analyses: associations between the Empirical Dietary Inflammatory Index (EDII) and cognition over 8 years, stratified by the presence of periodontitis and Helicobacter pylori (Hp) seropositivity in participants with unchanged eating habits.

Chou Y.‐C., Han M.‐L., Chiu T. H. T., et al., “Longitudinal Association of Inflammatory Diets on Cognition in Older Adults: Insights from the Oral–Gut–Brain Axis,” Journal of the American Geriatrics Society 73, no. 9 (2025): 2747–2756, 10.1111/jgs.19599.